Premium
Radon short range forecasting through time series preprocessing and neural network modeling
Author(s) -
Pasini Antonello,
Ameli Fabrizio
Publication year - 2003
Publication title -
geophysical research letters
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 2.007
H-Index - 273
eISSN - 1944-8007
pISSN - 0094-8276
DOI - 10.1029/2002gl016726
Subject(s) - artificial neural network , series (stratigraphy) , time series , residual , preprocessor , radon , data pre processing , range (aeronautics) , mixing (physics) , computer science , data mining , machine learning , algorithm , artificial intelligence , geology , engineering , physics , paleontology , quantum mechanics , aerospace engineering
In the framework of studies about the relevance of radon progeny measurements for the estimation of the mixing height, here a time series of radon data is analyzed and used for a short range forecasting activity. After a preprocessing of the time series in order to subtract the known periodicities, we perform forecasts of the future values of the residual series by means of neural network modeling. Finally we apply a simple box model to real data and forecast results, and obtain useful predictions of the mixing height during stability conditions.